On modern human origins and the emergence of complex language

Though much about the movements, migrations, interactions and material culture of early modern humans remains uncertain, rapid progress is being made by researchers. Questions concerning the non-material culture of our ancient ancestors, however – and, in particular, concerning their languages or modes of language-like communication – are far more problematic. What follows are a few reflections on what, in general terms, we know, and what the prospects might be for learning more.


Our ultimate African origins are not in dispute, but there are still fundamental disagreements between supporters of models which see modern humans as having migrated (more or less recently) to other continents, replacing other hominins in the process, and supporters of versions of a multiregional hypothesis who see the evolution of modern humans from earlier forms not just as an African but as a worldwide phenomenon involving significant interbreeding between different kinds of hominin, complex gene-flows and a number of regional continuities dating back at least 200,000 years.

Despite these disagreements it is, I think, becoming increasingly clear that the recent African origin model, the view that modern humans arose as a new species in Africa and migrated to other continents around 60,000 years ago, replacing existing human species in the process, is at best an oversimplification. For there is now firm genetic evidence that interbreeding occurred between modern humans and Neanderthals in Europe and between modern humans and Denisovans in Asia, as well as evidence that migrations of modern humans occurred more than 100,000 years ago. New versions of the 'out of Africa' model – which push back the dates of migrations and take into account interbreeding between different human groups – bring it closer to a multiregional model, though any consensus is still a long way off.

A recent University of Tübingen research project exemplifies how the African origin model is changing. The study focuses on modern humans who migrated east via the Arabian peninsula area where stone tools dating from more than 120,000 years ago have been found. Two significant migrations – a very early one (ca. 130,000 years ago) along the southern coast of the Arabian peninsula and a much later one via a northern route – were hypothesized, and the researchers' models predict in general terms the actual data (skull measurements and genetic data) of population groups currently living in the Asia-Pacific region.

According to Hugo Reyes-Centeno, a leading member of the Tübingen team, Aboriginal Australians, Papuans and Melanesians were "relatively isolated after dispersal along the southern route" and other Asian populations were largely descended from groups migrating much later (about 50,000 years ago) along other routes, the main one going via the north of the Arabian peninsula and northern Eurasia.

These results need to be treated with caution, however, as the data on which the models are based are necessarily extremely limited and incomplete. The results need also to be integrated with other data, including, for example, findings which indicate that Denisovans, who were widespread in Asia during the Late Pleistocene, contributed 4-6% their genetic material to present-day Melanesians.

The Denisovans were named after a cave in southern Siberia where a finger bone fragment from which DNA was able to be extracted was discovered. Geneticists have now managed to sequence the entire Denisovan genome to a high degree of accuracy.

Though closely related to Neanderthals, Denisovans seem to have interbred with an unidentified species and picked up some of their DNA. "Denisovans," claims David Reich of the Harvard Medical School in Boston, "harbour ancestry from an unknown archaic population, unrelated to Neanderthals." One possibility is that these scattered DNA fragments (which constitute only about 1% of the Denisovan genome) derive from H. heidelbergensis who lived in Europe and western Asia between about 600,000 and 250,000 years ago. Another possibility for the source of the archaic genes is Homo erectus.


While new archaeological and genetic evidence about the early history of humanity continues to accumulate and the broad outlines of a plausible story are beginning to fall into place, making progress in understanding linguistic (and many other cultural) factors will be difficult. Sure, archaeological findings may throw some light on questions concerning where and when complex languages first developed amongst human populations and also on the vexed question of whether Neanderthals used complex languages. For example, there is strong archaeological evidence that behavioral and cultural changes occurred amongst modern humans about 50,000 years ago, and this may well suggest that it was at about this time that human languages similar in structure and more or less equivalent in complexity to languages spoken today first appeared. Also, evidence of subtle genetic changes – relating to the FOXP2 gene, for example – may yield clues about which populations were capable of complex language and which were not.

Theories of culture-driven gene evolution tend to support the idea that humans developed language in a piecemeal but not necessarily always gradual process. The basic notion is that the existence of some form of primitive spoken language (without complex syntax or an extensive lexicon) may have created a cultural environment in which certain small genetic changes – e.g. in the FOXP2 gene which is important for (amongst other things) the fine motor control of vocalizations – could have had huge evolutionary advantages and so spread rapidly, prompting further cultural changes which in turn would have facilitated further genetic change, and so on.

But, in the final analysis, anatomical, genetic and broader archaeological findings will only ever be able to answer very general questions about culture and language (and only to a certain degree of probability) and it is difficult to see how specific questions concerning the nature of very early languages, or questions concerning to what extent particular groups such as Neanderthals or Denisovans developed their own languages or adopted (modified versions of?) the languages developed by modern humans, could move beyond the realm of speculation.

So, even if it could be argued convincingly on the basis of archaeological and genetic evidence that a certain population (modern or Neanderthal) at a certain time was extremely likely to have used a language of comparable complexity to today's human languages, the content of such a claim would necessarily be rather thin – and indeed linguistically vacuous – if any knowledge of the nature of that language is (and must remain) inaccessible to us.

In a recent post I referred to claims made by researchers at the Max Planck Institute that linguistic contacts with Neanderthals may have left discernible traces in the structure of non-African languages. However, given the time-frames involved and the fact that we only have access – and will only ever have access – to a minuscule fragment of the relevant linguistic data, it seems highly unlikely that even the most sophisticated computational approaches will be of much use. The researchers' claims are intriguing but, I would say, far too optimistic about what the sorts of approaches they are proposing could actually achieve.

It needs to be borne in mind that the earliest true writing systems for natural languages for which we have evidence date only from the 3rd millennium BC. Educated guesses and speculations about Proto-Indo-European, the hypothesized language from which the Indo-European language family (which includes Sanskrit, Persian, Latin, Greek and the Romance, Germanic and Slavic languages) is seen to derive, take us further back, but only to about 4000 BC.

Can knowledge of the languages spoken in recent times by preliterate peoples take us further back? Probably not. Though many of these languages have been recorded and analyzed, it would be a mistake to assume, even if the associated material cultures have been relatively stable for (in some cases) tens of thousands of years, that the languages themselves have exhibited anything like a similar stability.

The history of the human languages for which we have no historical written records (usually because there was no writing system but sometimes because written records have survived only in a fragmentary state or not at all) can only be hypothesized, largely on the basis of the elaborate comparative methods devised by philologists in the 19th and early-20th centuries coupled with general speculations about the speed and nature of linguistic change and its relation to broader social and cultural changes.

So it seems clear that, while broad evolutionary developments and migrations may eventually be able to be mapped with a high degree of confidence, the cultures of our ancient, preliterate forebears will only ever be able to be characterized in very general terms. Gaining substantive knowledge of the content of their cultures and belief systems, as of the actual (as distinct from the possible) nature and structures of the languages upon which these cultures were built and depended, lies forever beyond our grasp. The evidence just isn't there.

Neanderthal language

There has been a lot of talk in recent years about the intellectual and linguistic capacities of Neanderthals. This brief overview published last year in the research news section of the website of the Max Planck Institute reflects the currently popular view that their capacities were more or less equivalent to ours.

More controversial is the claim that the origins of modern language date not from about fifty or a hundred thousand years ago but from about one million years ago, "somewhere between the origins of our genus, Homo, some 1.8 million years ago and the emergence of Homo heidelbergensis." [Homo heidelbergensis is thought to represent our most recent common ancestor with Neanderthals, the split occurring about 500,000 years ago.]

More controversial still are claims being made by researchers at the Institute – and publicized in a recent New Scientist article [paywall] which is big on speculation but largely devoid of substantive content – that the cultural interactions between modern humans and their Neanderthal cousins included linguistic exchanges which left discernible traces in the syntax of non-African languages. Possible subtle structural differences between African and non-African languages coupled with detailed computer simulations of language spread would supposedly reveal something about the structural properties of hypothetical Neanderthal languages (which hypothetically impacted on non-African languages only). This is drawing a very long bow.

In a few days I will post some notes and reflections on some broader questions about early humans and the prospects for making progress in understanding their cultures and the nature of their languages.